agentleFS
appleweipingappleweiping/AGENT_RESOURCE/skills/vipin/feishu-bridge/SKILL.md

feishu-bridge

Skill78 starsChanged 24 days agoApache-2.0View on GitHub

Route Feishu/Lark content access for Codex. Use when the user asks to read, summarize, extract, search, create, update, or fill Feishu/Lark links, documents, wikis, Drive files, Base/bitable tables, Sheets, messages, forms, or pages; when Chinese prompts mention 飞书、飞书文档、飞书知识库、飞书多维表格、飞书表单、飞书链接、Lark, or lark-cli; or when deciding whether to use official Lark/Feishu API commands or Chrome automation for logged-in web pages.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

13 rules. About 0 are common, 2 are distinctive, and 11 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 11

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use the lark-cli binary the user installed.

    line 12
  2. Very specific:

    Read lark-shared/SKILL.md before configuring, logging in, switching identities, or handling scope errors.

    line 30
  3. Very specific:

    If the user gives a Feishu/Lark document or asks to read/summarize/update a doc, use lark-doc.

    line 31
  4. Very specific:

    If the user asks about a knowledge base / wiki node, use lark-wiki.

    line 32
  5. Very specific:

    If the user asks to search, upload, download, find a file, manage comments, or discover cloud objects, use lark-drive first.

    line 33
  6. Very specific:

    If a document contains an embedded spreadsheet or bitable, extract the token and switch to lark-sheets or lark-base rather than stopping at …

    line 34
  7. Very specific:

    If the user asks to read/write a multi-dimensional table / 多维表格 / bitable, use lark-base.

    line 35
  8. Very specific:

    If the user asks to send, reply, search, or inspect messages, use lark-im, and dry-run risky sends before execution.

    line 36
  9. Very specific:

    If the user gives a web link that cannot be cleanly handled by token/API, or asks to fill fields/buttons on a Feishu page/form, use browser …

    line 37
  10. Very specific:

    When config init or auth login emits a verification URL, forward the raw URL exactly as returned.

    line 51
  11. Distinctive:

    Do not expose app secrets, access tokens, or private document content in public wiki pages.

    line 55
  12. Very specific:

    Use --dry-run before high-risk writes when available.

    line 57
  13. Distinctive:

    Read/write tests are performed only against user-approved Feishu resources.

    line 65

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Feishu Bridge

Use this as the router for Feishu/Lark work. Prefer official lark-cli API access whenever a task can be expressed as Docs/Wiki/Drive/Base/Sheets/IM/Calendar/etc. Use browser automation only for web-only flows such as opening arbitrary shared links, pages that require the user's live browser login state, or filling interactive forms.

Local Runtime

Use the lark-cli binary the user installed. If $env:LARK_CLI is set, it holds the full path to the binary; otherwise expect lark-cli on PATH:

$LARK_CLI = if ($env:LARK_CLI) { $env:LARK_CLI } else { "lark-cli" }
& $LARK_CLI --version

I use lark-cli 1.0.32 from the larksuite/cli release v1.0.32; see references/local-runtime.md.

For browser automation I use the chrome-automation skill from anbeime/skill, which drives the user's logged-in Chrome over CDP. It is not part of this repo; if it is not installed, use whatever browser-automation skill or tool the agent has. If there is none, say that the web-only step is blocked instead of guessing.

Routing

  1. Read lark-shared/SKILL.md before configuring, logging in, switching identities, or handling scope errors.
  2. If the user gives a Feishu/Lark document or asks to read/summarize/update a doc, use lark-doc. For docs v2 commands, always pass --api-version v2.
  3. If the user asks about a knowledge base / wiki node, use lark-wiki.
  4. If the user asks to search, upload, download, find a file, manage comments, or discover cloud objects, use lark-drive first.
  5. If a document contains an embedded spreadsheet or bitable, extract the token and switch to lark-sheets or lark-base rather than stopping at the embedded tag.
  6. If the user asks to read/write a multi-dimensional table / 多维表格 / bitable, use lark-base.
  7. If the user asks to send, reply, search, or inspect messages, use lark-im, and dry-run risky sends before execution.
  8. If the user gives a web link that cannot be cleanly handled by token/API, or asks to fill fields/buttons on a Feishu page/form, use browser automation against the user's logged-in browser (chrome-automation if installed).

Authentication

Assume personal user OAuth by default. Use bot identity only when the user explicitly asks for bot/app behavior.

Typical setup:

& $LARK_CLI config init --new
& $LARK_CLI auth login --recommend
& $LARK_CLI auth status

When config init or auth login emits a verification URL, forward the raw URL exactly as returned. Do not re-encode or rewrite it. If OAuth is not completed in the current session, report API read/write tests as blocked by user authorization rather than claiming success.

Safety

  • Do not expose app secrets, access tokens, or private document content in public wiki pages.
  • For public wiki records, preserve only neutral metadata, capability notes, and smoke-test status unless the user explicitly permits more detail.
  • Use --dry-run before high-risk writes when available.
  • Confirm intent before deleting, bulk editing, sending messages, changing permissions, or modifying important business data.

Smoke Test Checklist

  • & $LARK_CLI --version works.
  • Domain help works, for example docs --api-version v2 --help, wiki --help, base --help, and drive --help.
  • OAuth/config status is checked.
  • Read/write tests are performed only against user-approved Feishu resources.
  • Browser-fill tests use the browser-automation skill on a test page or explicitly approved Feishu form.
  • Tell the user which capabilities were verified and which are blocked (for example by missing OAuth scopes).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.